2. By the end of this session, students should be able to:
ī§ Distinguish between experimental and observational
studies.
ī§ Identify the design of a particular study.
ī§ Discuss the factors that determine when a particular
design is indicated.
ī§ Identify the strength and limitation of each studies
2
3. What is Study Design?
Is a specific plan or protocol for conducting
the study, which allows the investigator to
translate the conceptual hypothesis into an
operational one.
Arrangement of conditions for the collection &
analysis of data
Logical model that guides the investigator in the
various stages of the research process
Overall structure of the study
3
4. Epidemiologic Study Designs
ī§ The basis for the study designs is the distinction between
Descriptive epidemiology and Analytic epidemiology
Descriptive Epidemiology: seeks to measure the frequency
in which diseases occur or collect descriptive data on possible
causal factors.
Analytic Epidemiology: attempts to specify in more detail
the causes of a particular disease
Apart from the simplest descriptive studies, almost all
epidemiological studies are analytical in character. Pure
descriptive studies are rare
4
5. Observational and Experimental Study Designs
ī§ Epidemiological studies can be classified as either
observational or experimental.
Observational Studies
ī§ Allow nature to take its course: the investigator
measures but does not intervene.
ī§ They include studies that can be called descriptive or
analytical
Experimental/interventional Studies
ī§ Involve an active attempt to change a disease
determinant such as an exposure or a behaviour or the
progress of a disease through treatment
5
7. Does the study test hypothesis?
Does the study have comparison groups?
Is the study unit individual?
Does the researcher intervene the
natural course of action?
No
Descriptive
Yes
Analytical
No
Correlational/
Ecological
Yes
Case report
Case series
Cross sectional/
Prevalence studies
No
Observational
Æ
Cohort
Æ
Case-Control
Cross-Sectional
Yes
Intervention/Experimental
RCT
Field trials
Community trials 7
9. Study Design Sequence
Hypothesis Formation
Hypothesis testing
Case
Report
Analytic
Study
Descriptiv
e Study
Case
Series
Cohort
Case-
Control
Cross-
Sectional
Animal
Study
Lab.
Study
Clinical
Trials
9
11. Features of descriptive study designs
ī§ Purpose: Mainly concerned with the amount and
distribution of diseases with respect to time, place and
person within a population.
ī§ In other words âĻ. To identify health problems and
patterns of disease that exist.
ī§ Descriptive studies generally precede analytic studies
designed to investigate determinants of disease.
11
12. Features of Descriptive Study
ī§ Useful for health managers to allocate resource and to
plan effective prevention programmes.
ī§ Useful to generate epidemiological hypothesis in the
search for disease determinants or risk factors.
ī§ Inexpensive and less time-consuming: can use
information collected routinely.
ī§ Most common type of epidemiological study in the
medical literature.
12
13. Descriptive Studies Cont.
ī§ Data source is routinely available data or data
obtained in special surveys.
ī§ No attempt to gather data on controls; meaning no
comparison group
ī§ Pure descriptive study design do not give opportunity
to make an attempt to analyse the links between
exposure and effect.
13
14. Types of descriptive studies
1. Population as study subject
īąCorrelational /Ecological studies
2. Individual as study subjects
īą Case report
īąCase series
īą Cross-sectional surveys
14
15. Ecological/Correlational Studies
ī§ Measures that represent characteristics of entire
populations
ī§ Used to describe disease and to postulate causal
associations (Useful for generating hypotheses).
ī§ Usually rely on data collected for other purposes;
15
16. Ecological Studies âĻ
ī§ The two key features that distinguish a traditional
ecologic study from other types of epidemiologic
studies are:
1. The population unit of analysis
2. An exposure status that is the property of the
population
16
17. Ecological Cont. . .
Strengths:
Simple to conduct and thus attractive
Cheap and quick(use of secondary data)
Data can be used from populations with widely differing
characteristics or extracted from different data sources.
Limitations:
Uses average exposure levels rather than actual levels of
exposure
Difficult to interpret, since it is seldom possible to examine
directly the various potential explanations for findings.
Since the unit of analysis is a group, the link between
exposure and effect at the individual level can not be made.
17
18. Ecological contâĻ
ī§ Some investigations cannot be classified as traditional
ecologic studies because they have both ecologic and
individual-level components
ī§ Consider, for example, a study conducted in Norway to
determine if chlorinated drinking water was associated
with the occurrence of birth defects
ī§ The study used group-level data on the exposure and
individual-level data on the birth defects and
confounding variables.
ī§ Such studies are considered partially ecologic
18
19. Ecological fallacy/Bias
Results if inappropriate conclusions are drawn on the
basis of ecological data.
Occurs because the association observed between
variables at the group level does not necessarily
represent the association at the individual level
19
20. Example
ī§ In 1951, Durkheim, reported an association
between the rate of suicide and proportion of
Protestant religion.
ī§ How might the ecologic fallacy be responsible for
this association?
20
21. Group exercise
ī§ Form a group with 3 members
īē Identify one research idea that can be conducted by using
Ecological study design and answer the following
questions(5-10min.)
ī What will be the sources of information ?
ī What is the significance/importance of the study?
ī What will be the limitation of the study ?
īē Present your work to other groups (5min. for each group)
21
22. Case Reports
ī ⥠Careful and detailed report by one or more
clinicians of the profile of a single patient
ī Document unusual medical occurrences
ī Can provide clues in identification of a new disease
or adverse effects of exposures
Example
īē One case of pulmonary embolism observed 5 weeks after Oral
Contraceptive usage â clue for the association of OC use and
venous thromboembolism
22
23. Case Reports
Limitations:
ī§ No appropriate comparison group
ī§ Cannot be used to test for presence of a valid
statistical association
ī§ Since based on the experience of one person:
īē presence of any risk factor may be purely coincidental
īē not a true epidemiologic design
23
24. Case Series
ī§ âĄ
Description of clinical/epidemiologic
characteristics of a number of patients with a given
disease
ī§ Collection of individual case reports occurring
within a fairly short period of time
Example
In 1974, Creech and Johnson reported a case series
of three men with angiosarcoma of the liver among
workers at a vinyl chloride plant
24
25. Case Series
Strengths:
ī§ Used as an early means to identify the beginning or
presence of an epidemic
ī§ Can suggest the emergence of a new disease (i.e.
AIDS)
Limitations:
ī§ Lack of an appropriate comparison group
ī§ Cannot be used to test for presence of a valid
statistical association
25
26. Individual Exercise
ī§ Compare and contrast Case report and Case series
by the following condition
1. The Statistical test to be used
2. Time need to conduct
3. The contribution for discovery of new disease and
4. Development of hypothesis
26
27. Cross-sectional study
Examines the relationship between diseases (or other
health related characteristics) and other variables of
interest as they exist in a defined population at one
particular time
Populations are commonly selected without regard to
exposure or disease status
Carried out for public health planning & for etiologic
research
27
28. Cross-sectional study contâĻ.
ī§ For each subject, exposure and disease outcome are assessed
simultaneously (hence also called a "prevalence
study/surveyâ)
ī§ It is useful for raising questions of the presence of
association rather than for testing hypothesis
ī§ But it provides evidence of association for factors that
remain unaltered
ī§ Most useful for conditions that are:
īŧ Not rapidly fatal,
īŧNot terribly rare, and/or
īŧNot routinely brought to medical attention
28
29. Cross-sectional study contâĻ.
ī§ Usually measure disease prevalence in relation to
exposure prevalence
īē i.e current disease status is usually examined in relation to
current exposure level
ī§ It is also possible to examine disease prevalence in
relation to past exposure if the dates of the exposure
are ascertained
29
30. Advantages
ī§ Provides prevalence estimates of exposure and disease
for a well defined population
ī§ When they are based on a sample of the general
population, their results are highly generalizable
ī§ Can be carried out in a relatively short period of time
with low cost
ī§ Can evaluate multiple risk (and protective) factors and
health outcomes at the same point in time
30
31. Advantages
ī§ One-stop, one-time collection of data
ī§ Provide much information useful for planning health
services and medical programs
ī§ Show relative distribution of conditions, disease,
injury and disability in groups and populations
ī§ Studies are based on a sample - do not rely on
individuals that present themselves for medical
treatment
31
32. Disadvantages
1. Are not able to infer the temporal sequence between
the exposure & disease
i.e. Which came first, chicken or the egg?
īē This occurs when exposure under study is changeable
characteristic such as place of residence, habit such as
drinking, smoking and physical exercise
Information on all factors is collected simultaneously,
so it can be difficult to establish a putative "cause'
preceding the "effect'.
32
33. Disadvantages ContâĻ
The temporal inference problem can be avoided if
An inalterable/unchangeable characteristic is
the focus of the investigation. Eg genetic trait
The exposure measure reflects not only
present but also past exposure.
33
34. Disadvantages contâĻ
2. Length-biased sampling :-The cases identified
will over represent cases with long duration
(prevalent cases) and under represent those with
short duration of illness.
ī People who die soon after diagnosis or who recover quickly
are less likely to be identified as diseased.
ī This can bias the results if duration of disease is associated
with the exposure under study
34
35. Disadvantages contâĻ
ī§ If the exposure does not alter disease risk but causes the
disease to be mild and prolonged when contracted (so
that the exposure is positively associated with duration),
the prevalence of the exposure will be elevated among
cases.
ī§ As a result, the exposure-disease association will be
observed in a cross-sectional study, even though
exposure has no effect on disease risk and would be
beneficial if disease occurs.
35
36. Disadvantages contâĻ
Cross sectional
study
Real cause Disease /No exposure (recover or die) âĻâĻâĻ
(non-exposed cases will not remain diseased for longer duration)
Real cause Disease/Exposure Exposed cases survive longer
Result:
īē Exposure affect the duration of the disease but not the risk of developing disease
īē Many exposed cases available for cross sectional study
īē Large proportion of people with the disease are found to be exposed
īē False positive association
36
37. Disadvantages contâĻ
ī§ If the exposure does not alter disease risk but causes the
disease to be rapidly fatal if it is contracted (so that
exposure is negatively associated with duration), then
prevalence of exposure will be very low among cases.
ī§ As a result, the exposure-disease association observed
in the cross-sectional study will be negative, even
though exposure has no effect on disease risk and
would be detrimental if disease occurs
37
38. Disadvantages contâĻ
Cross sectional study
Real cause disease /exposure (recover or die âĻâĻâĻâĻ
(exposed cases will not remain diseased for longer duration)
Real cause Disease/no exposure Non-exposed cases survive longer
Result:
â Exposure affect the duration of the disease (decrease) but not the risk of
developing disease
â Many non-exposed cases available for cross sectional study
â Less proportion of people with the disease are found to be exposed
â False negative association
38
39. Disadvantages contâĻ
3. Healthy Worker Survivor Effect
In studies conducted in occupational settings,
because these studies include only current and not
former workers, the results may be influenced by
the selective departure of sick individuals from the
workforce
īē Those who remain employed tend to be healthier than
those who leave employment.
īē This phenomenon known as the âhealthy worker survivor
effect,â generally attenuates an adverse effect of an
exposure
39
40. Disadvantages contâĻ
It may not show strong cause-effect relationships if
sample size is small.
Susceptible to selection bias (e.g. selective
survival)
Susceptible to misclassification (e.g. recall)
Not good for rare diseases or rare exposures
40
41. Analysis
ī§ Either compare prevalence rate of the outcome in exposed Vs
non-exposed, or
ī§ Compare prevalence rate of the exposure in those with and
without the outcome
ī§ Timing of the subdivision of the study population into comparison
groups distinguishes cross sectional studies from other
observational analytic studies
ī§ In cohort and case control studies, this takes place prior to the data
collection process
ī§ In a cross sectional study, this takes place after the information
has been collected
41
42. Example
ī§ compare prevalence rate of the outcome in
exposed Vs non-exposed,
Disease Status Total
Disease Not
Diseased
Exposer
Status
Exposed
a b a+ b
Non-
Exposed
c d c+d
Total a+c b +d a+c+b +d
42
43. Example
Disease Status Total
CHD No CHD
Cholesterol
level
High
100 400 500
Normal/low
50 450 500
Total 150 850 1000
Interpretation:
In this study population, the prevalence of CHD is 2
times higher among those with high cholesterol,
compared to the prevalence in those with normal or low
cholesterol. 43
44. Discussion Point
ī§ How do you conduct cross sectional study to
assess whether khat chewing causes
psychosis
īē Who will be the study units? ; how you will select
them?
īē How and when do you conduct data collection?
īē What will be the limitation of the study?
44
46. 2. Analytic Study Designs
ī§ Focus on the determinants (causes) of
diseases.
ī§ Used to test hypothesis
ī§ Major distinguishing feature of analytic
studies is the use of controls.
ī§ The comparison is explicit, since the
investigators assembles groups of individuals
for the specific purpose of systematically
determining whether or not the risk of disease
is different for individuals exposed or not
exposed to a factor of interest
46
47. Types of analytic studies
A. Observational analytic studies
īą Case control
īą Cohort
īą Cross sectional?
B. Intervention/Experimental studies
īą Randomized controlled trials
īą Field trials
īą Community trials
47
48. Also known as Case-referent study
This approach begun to developed in mid. 20th
c. as part as response to needs that
accompanied the shift from acute to chronic
disease
Subjects are selected on the basis of whether
they do (cases) or do not (controls) have a
particular disease under study; from same
population.
48
49. Case-Control Design
Groups are compared with respect to
the proportion having a history of an
exposure or characteristics of interest
Specifically good for studying rare
diseases & diseases with very long
latency periods
Several risk factors/ exposures can
be studied
49
51. Case-Control Design cont..
Outcome in case
control study
ī§ The outcome has
traditionally been the
presence or
absence of
disease.
ī§ However, other
outcomes can also
be studied
e.g. disability, smoking
cessation etc
51
52. ī§ The case-control method (especially the
analysis) was developed in the 1950s as
an approach to the problem of
investigating risk factors for diseases with
long latent periods andrare diseases.
52
54. Types of Case Control Study
1. Retrospective case control study: all the cases
of the disease have been diagnosed at the
time the investigators initiates the study.
2. Prospective case control study: the study is
begun and all new cases that are diagnosed
within the next period of time are included.
Thus in the context of these definitions the use
of the term retrospective to refer to all case-
control studies is inappropriate.
54
55. Definition and Selection of cases
ī§ Establish strict diagnostic criteria for the
disease.
ī§ Setting clear definition of cases
īē It is often useful to perform analyses separately
for cases classified as definite, probable or possible
ī§ Representing spectrum of disease: mild,
moderate and severe groups
ī§ Incident (newly diagnosed) Vs Prevalent
(existing at a point in time) cases
55
56. Selection of cases contâĻ
ī§ To study the causes of disease incident cases are
preferable
īē Because usually the interest is in knowing the factors that lead
to developing a disease rather than factors that affect its
duration
ī§ The reason is that any risk factors we might identify in
a study using prevalent cases may be related more to
survival with the disease than to the development of
the disease
īē Thus the risk identified will not be the characteristics of all
patients rather the survivors only
56
57. Selection of cases contâĻ
ī§ Some times epidemiologists have no choice
but to rely on prevalent cases (esp. when
the exact onset is difficult to pinpoint)
īē Studies using prevalent cases must be
interpreted cautiously
ī§ It is impossible to determine if the
exposure is related to the inception of the
disease, its duration, or a combination of
the two.
57
58. Sources of Cases
Hospital Vs General Population Cases
Hospital Based: Population Based
âĸEasy and inexpensive to
Conduct
âĸIt is desirable to select cases
from several hospitals
âĸProne to selection bias.
âĸAvoids selection bias
âĸInvolves locating and obtaining data from a
individuals or a random sample from a def
population
âĸAllows the description of a disease in the
entire Population.
âĸDirect calculation of rates possible
Incident Vs Prevalent cases
Prevalent Cases Incident case
īIncrease sample size available for
rare disease.
īDifficult to establish temporal
sequence between exposure
and outcome reverse causation.
īHelpful to establish temporal
relationship between exposure and
outcome.
īRecords are easily obtainable and 58
59. Selection of controls
Involves consideration of a number of issues
including :
īŧThe characteristics and source of the cases
īŧThe need to obtain comparable
information from cases and controls
īŧ Practical and economic considerations.
The control subjects should be selected to be
comparable to the cases.
59
60. Selection of controls cont..
ī§ Controls should have the same exposure distribution as
the source population from which cases are drawn.
ī§ The controls should be similar to the cases in all respects
other than having the disease in question or Controls
should be selected independent of exposure.
ī§ Whether they should be representative of all persons without
the disease in the population from which the cases are
selected
ī§ probability of selection is proportional to time spent in
the source population same probability of being diagnosed
as a case at the time of selection.
60
61. Hospital Controls
Advantages: Disadvantage
īąEasily identified and readily
available in sufficient number with
reduced cost.
īąMore likely than healthy individuals
to be aware of antecedent exposures
or events- minimize recall bias
īąControls are also likely to have
been subject to the same intangible
selection factors -minimize
selection bias
īąMore likely to be cooperative
reduce bias due to non-response
īąIll individuals are
different from healthy
īąDanger of altering the
direction of
association or masking
a true association
between exposure and
outcome
61
62. General Population Controls
Advantages: Disadvantage
īąGeneralizable
īąGood when cases are
selected to represent
affected individuals in a
defined population.
īąCostly and time-consuming
īąâĄ
Recall bias - controls may
not recall exposures with the
same level of accuracy.
īąâĄ
People might be less
motivated to participate for the
same reason given above,
which increases non-
response rate, i.e., selection
bias.
62
63. Special controls
Consists of special groups such as friends, neighbor,
relatives, or spouses of cases
Advantage:
âĸThey are healthy
âĸMore likely to be cooperative
âĸ Offer a degree of control of important confounding factors
Limitations
âĸIf the study factor itself is one for which family members and
friends are likely to be similar to the cases,
âĸan underestimate of the true effect of the exposure of
interest may result.
63
64. How many control groups?
īIdeally a single control group
īwhen the number of available cases and controls is
large and the cost of obtaining information from both
groups is comparable, the optimal control to-case ratio
is 1:1
īHowever, it is often difficult, especially with
hospitalized controls.
īThe use of multiple control groups is also indicated
when there is concern that one selected group has a
specific deficiency that could be overcome by the
inclusion of another control group.
64
65. How many control groups? Cont..
âĸ As the number of controls per case increases, the
power of the study also increases.
âĸ not generally recommended that this ratio increase
beyond 4:1
âĸ When the entire population of potential eligible controls
is known, a random sample of the required sample can
be chosen
âĸ regardless of the specific method of selection employed,
it is important to follow clearly defined, objective, and
reproducible procedures.
65
66. Ascertainment of disease and exposure status
ī§ After the case and control series have been
defined in terms of characteristics and sources
ī§ Information of the disease and exposure must be
obtained
ī§ Any potential source of information must be
carefully considered
i.e. the ability to provide accurate as well as
comparable information for all study groups
66
67. Sources of information
Disease status
âĸ Review of death certificates, case registries that
maintain ongoing surveillance
âĸ Hospital admission or discharge records
âĸ Pathology department log books
Exposure status
âĸ The study subject themselves
âĸ Surrogate (eg. Mother of the child, spouse)
âĸ Medical records
67
68. Analysis of Case Control Results
ī§ The analysis of case control study is basically a
comparison between cases and controls with respect to
the frequency of exposure
ī§ Odds ratio: the ratio of the odds that the cases were
exposed to the odds that the controls were exposed
ī§ Rates of disease for those exposed and non exposed
can be computed and compared directly
If the case control study is population-based, or if
estimates of disease incidence are available from an
out side source
68
69. The odds ratio in case control and cohort studies
Cohort Study
Developed
Disease
Do Not Develop Disease Total
Exposed a b
Not Exposed c d
Case-Control Study
Cases Control Total
Were Exposed a b
Were Not
Exposed
c d
69
70. Analysis
ī§ Although the underlying conditional probability
relationship defined OR differs for the case-control
and cohort study designs, the same simplified
calculation formula can be used in both cases
ī§ When is the OR a good estimate of the RR?
1. When the cases studied are representative, with
regard to exposure, of all people with the disease in
the population from which the cases were drawn.
2. When the controls studied are representative,
with regard to exposure, of all people without the
disease in the population from which the cases were
drawn.
70
71. Analysis
3. When the disease being studied occurs
infrequently
ī§Very few people will develop the disease in an exposed
population (a very small compared to b, a+b can
approximate as b)
ī§Similarly, very few non exposed person develop the
disease, (c very small compared to d, c+d approximate as
d)
Therefore, we may calculate the relative risk as
which is the odds ratio
71
73. When is it desirable to use case control?
Case control study is preferable to cohort study in five
types of situations:
1. When the exposure data are difficult or
expensive to obtain
2. When the disease is rare
3. When the disease has a long induction/ latent
period particularly for prospective
4. When little is known about the disease
5. When the population under study is dynamic
73
74. Applications of Case Control Studies
1) For determining the causes of disease
2) For problem solving activities within the
practice of public health and medicine:
īē E.gs- evaluating the effectiveness of vaccines
īē evaluating treatment and prevention programs
īē Investigating outbreaks of disease
74
75. Advantage of Case-Control
ī Relatively cheap/quick
ī Can investigate a number of risk factors
ī Useful for rare diseases
ī No loss to follow up
75
76. Disadvantage of Case-Control
Vulnerable to bias in selection of cases and
controls (sampling/selection bias)
īļ(Primary challenge in case-control study is the
identification of the appropriate âstudy baseâ from
which to select controls)
Vulnerable to bias in measurements (recall
bias)
Not suitable for rare exposures
Does not establish the sequence of events
76
77. Bias in Case-Control
1. Selection Bias:- Ideally, the sample of cases would be
a random sample of everyone with the disease butâĻ
Sampling in case-control studies frequently miss many
eligible cases
No medical attention sought
Attention sought elsewhere
Misdiagnosed cases
Deaths
2. Recall Bias: -Occurs from retrospective nature of data
collection
Systematic differences in recollection of exposures
between cases and controls
Cases usually more likely to remember
Cases missed
from being
included in the study
77
78. Controlling confounding
âĸ Controlling confounding in analysis
â by stratifying on confounding factor
â by multiple regression
âĸ Controlling confounding in methods
â by matching
â by restriction
78
79. Case-Control: Matching
âĸ Random sample may not be possible
âĸ Matching can improve efficiency of study
â Useful when cases are scarce
â Useful when sample size is small
âĸ Can control for confounding due to factors
that are difficult to measure
79
80. Quiz 2 five point.
1. Discuss the advantage of case control
study. At list three.
2. List at list three situations of Case
control study is preferable to cohort
study.
3. Define the meaning of retrospective
and prospective case control study.
80
81. What is cohort ?
ī§ Group of individuals
Examples
īē birth cohort
īē cohort of guests at wedding reception
īē occupational cohort of chemical plant workers
īē cohort from the general population
Cohort study
Group of individuals followed up for a specified
period of time with assessment outcomes
(diseases/deaths)
81
82. Cohort studies
īHealthy subjects are defined according to their exposure
status and followed over time to see the incidence of disease
or death
īOther expressions of cohort study:
īē Follow up study
īē Incidence study
īē Longitudinal study
ī§ Cohort membership
1. Being at risk of outcome(s) studied
ī Protected (e.g. immunized)
ī Induction time, latency
2. Being alive and free of outcome at start of follow-up
ī Occurrence of same disease more than once in same individual ?
82
83. Method: Cohort study
Identify group of
ī Exposed subjects
ī Unexposed subjects
Follow up for disease occurrence
Measure incidence of disease
Compare incidence between exposed and
unexposed group
83
84. Cohort study: objective
84
The cohort study design objective is to
compare:
ī§ an incidence rate in an exposed
population
ī§ to the rate that would have been observed
īē in the same population, at the same time
īē if it had not been exposed
85. Information on exposure
īąQuestionnaires
âĸ Interview
âĸ Postal
īąClinical examination
âĸ Nutritional status
âĸ Height and weight
âĸ Blood tests (E.g. Cholesterol, antibodies, HIV)
īąRegistered data
âĸ Demographic data
âĸ Occupation, Income, Soc.Ec.Class, Ethnic group, Religion
âĸ Patient journals (Treatments, Smoking habits, Drugs, X-rays)
īąEnvironmental information (Groups,
Individuals)
85
86. Information on outcome
īē Questionnaires
ī Interview (verbal autopsy)
ī Postal
īē Clinical examination
ī Biological tests
īē Registered data
ī Death register/Death certificates/Autopsies
ī Hospital discharge registries
ī Registries of diseases and injuries
ī Cancer registries
86
88. The prospective cohort design
ī§ Investigator starts with a group of individuals
apparently free from disease
ī§ Participants are grouped on the basis of past
or current exposure and are followed into the
future in order to observe the outcome of
interest
ī§ Followed through time to determine incidence
of disease among exposed and the
unexposed (or at different levels of exposure)
88
89. Prospective cohort study design
89
Population
Sample
Risk factor
present
Disease
THE FUTURE
THE PRESENT
Risk factor
absent Disease
No
disease
No
disease
91. Prospective cohort study contâĻ
âĸ Plan beforehand
â What data to collect, when, and by which
method
âĸ Study must continue over a long period of time in
order to observe a sufficient number of cases
âĸ Length of follow-up time required is dependent on
īąThe incidence rate and
īąThe size of the population at risk.
91
92. Retrospective Cohort Study
īą Both the exposures & outcomes have already occurred
when the study begins
īą Studies only prior outcomes & not future ones
īą Usually less costly than prospective cohort studies and
also take less time to complete.
īą Limited to studies of outcomes which have been
documented in so far as data are retrievable and
reliable.
īą Especially suitable for studies of rare exposures, or
where the latent period between exposure and disease
is long.
92
94. Retrospective Cohort Study
ī§ Identify cohort in the past
īē E.g., through records or administrative
database
ī§ Determine exposure or prognostic factors in
the past
īē Again through records or databases
ī§ Identify outcome
īē Outcome can be identified in past or present
īē Time sequence: Outcome must be after
exposure
94
95. Ambidirectional Cohort Design
ī§ Has both prospective & retrospective
components
ī§ Data are collected both retrospectively and
prospectively on the same cohort
īē Identify cohort in the past
īē Determine exposure/prognostic factors in the past,
present or future
īē Identify outcome in the past, present and in the
future
ī§ Is most useful for exposures having both short term
and long term effects
95
96. Which type of cohort study should be used?
The decision to conduct a retrospective,
prospective, or ambidirectional study
depends on:
ī The research question
ī Practical constraints such as time &
money
ī Availability of suitable study
population and records
96
97. Advantage Disadvantage
Retrospectiv
e cohort
īŧMore efficient
īŧfor investigating
diseases that take a long
time to develop
īŧMinimal information
is usually available on
the exposure,
outcome, and other
key variables
Prospective
Cohort
īŧMore detailed
information
īŧFollow up may be easier
īŧless vulnerable to bias
īŧMore reliable than the
retrospective
īŧ Expensive time
consuming
īŧMay not be
appropriate for
diseases with long
induction period
97
98. Types of Population Studied
Open or dynamic population
ī§ Individuals may enter or leave at any time
because its membership is defined by a
changeable characteristics
ī§ E.g smoking, living in a specific geographic area
ī§ Such cohort studies conducted in an open
population usually take into account population
changes such as in- and out- migration
ī§ The person-time data could be used as the
denominators for the incidence rates
98
99. Types of population studied contâĻ
Fixed cohort
ī§ Is defined by an irrevocable event (e.g
eating contaminated food)
ī§ Groups are followed from a defined starting
point (usually marked by the event) to a
defined ending point
ī§ Incidence rates are the appropriate
measure of disease frequency when the
population experiences losses to follow up
99
100. Types of population studied contâĻ
Closed cohort
ī§ Is defined by an irrevocable event
ī§ Has defined starting & ending points for
follow up
ī§ Has no losses to follow-up
ī§ Cumulative incidence or average risk is used
as a measure of disease frequency because
there are no losses to follow-up
100
101. Selection of the Exposed Population
âĸThe exposed population should relate to the
hypothesis:
âĸFor common exposures (e.g. smoking, coffee
drinking) and relatively common chronic
diseases, the general
population/geographically-defined areas are
good choices.
âĸFor rare exposures, âspecial cohortsâ are
more desirable (e.g. particular occupations or
environmental factors in specific geographic
locations).
101
102. Selection of the Exposed PopulationâĻ
âĸ Although cohort studies are not optimal for
evaluation of rare diseases, certain
outcomes may be sufficiently common in
âspecial exposure cohortsâ to yield an
adequate number of cases.
âĸ To enhance validity, some exposed
populations are selected for their ability to
facilitate complete and accurate information
(e.g. doctors, nurses, entire companies,
etc.).
102
103. Selection of the Comparison Group
âĸ The groups being compared should be
as similar as possible on all factors that
relate to disease other than the
exposure under investigation
âĸ (e.g. to reduce the potential for
confounding).
ī§ Ability to collect adequate information
from the non-exposed group is
essential.
103
104. Selection of the Comparison GroupâĻ
âĸ Internal Comparison Group:
âĸ Members of a single general cohort are
classified into exposed and non-exposed
categories.
âĸ Most often used for common exposures.
âĸ The non-exposed group becomes the
comparison group.
âĸ Must be careful of other potential
differences between the exposed and
non-exposed groups.
104
105. Selection of the Comparison GroupâĻ
âĸ General Population Comparison Group:
âĸ The general population will probably
include some exposed persons.
âĸ Due to the âhealthy worker effect,â the
general population may be expected to
experience higher mortality than most
occupational cohorts.
âĸ Comparisons with population rates are
possible only for outcomes for which
population rates are available.
105
106. Selection of the Comparison GroupâĻ
âĸ Special Exposure Comparison Group:
âĸ Another cohort with demographic
characteristics similar to the exposed
group, but considered non-exposed
to the factor of interest is selected
(e.g. another occupational group).
âĸ Note: To enhance validity, it may be
important to have multiple
comparison groups.
106
107. Sources of Exposure Information
âĸ Pre-existing Records:
Advantages:
--- Inexpensive
--- Relatively easy to work with
--- Usually unbiased since the data were
collected for non-study purposes
107
108. Sources of Exposure InformationâĻ
âĸ Pre-existing Records:
ī§ Disadvantages:
īē Exposure information may not be
precise enough to address the
research question.
īē Records frequently do not contain
data on potential confounding
factors.
108
109. Sources of Exposure InformationâĻ
âĸ Self Report (interviews, surveys, etc.)
Advantages:
--- Opportunity to question subjects on
as many factors as necessary.
--- Good for collecting information on
exposures not routinely recorded.
109
110. Sources of Exposure InformationâĻ
âĸ Self Report (interviews, surveys, etc.)
Disadvantages:
--- Subject to response bias (e.g. due to
stigma, response expectations, etc.).
--- Subject to interviewer bias.
--- Subjects may be sufficiently unaware
of their exposure status (e.g.
chemical exposure).
110
111. Sources of Exposure InformationâĻ
âĸ Direct Measurement
âĸ If obtained in a comparable manner, can
provide objective and unbiased exposure
ascertainment (e.g. blood pressure, serum
samples, environmental measurements, etc.).
âĸCan be used on a fraction of the cohort to
validate other types of exposure
ascertainment.
111
112. Sources of Exposure InformationâĻ
âĸ Repeated Measurements
âĸ If frequency of exposure changes over
follow-up, repeated measurements allows
for revision of exposure classification.
âĸ Periodic questioning of cohort members
allows for newly identified exposures of
interest to be measured.
--- Good for âtransientâ exposures.
112
113. Sources of Outcome Information
âĸ Death certificates (National Death Index) âfor
some causes, notoriously unreliable
âĸ Clinical history
âĸ Self-reports
âĸ Medical examination (periodic re- examination
of the cohort)
âĸ Hospital discharge logs
113
114. Outcome Information
âĸ Procedures for identifying outcomes must be
equally applied to all exposed and non-exposed
individuals.
âĸ Goal is to obtain complete, comparable, and
unbiased information on the health experience of
each study subject.
âĸCombinations of various sources of outcome
data may be necessary.
114
115. Population based cohort studies
Strengths
ī§ Accurate measurement
of exposure (but often
measured only once)
ī§ Valuable for studying
risk factors of fatal
diseases
ī§ Can study several risk
factors
ī§ Can study several
outcomes
Weaknesses
ī§ Expensive and
inefficient in areas that
lacks population based
registries for outcome
assessment
ī§ Unsuitable for rare
diseases
ī§ Unsuitable for rare
exposures
115
116. 116
Historical (Retrospective) Cohort Studies
Strengths:
âĸ cohort easier to assemble (inception period in past)
âĸ baseline measurements already available
âĸ follow-up period already taken place
âĸ less costly and time-consuming
Weaknesses:
âĸ no control over the quality of past measurements
âĸ incomplete data sets
âĸ control for confounding may be incomplete